HGR 2.0： A New Rule Induction Algorithm

This paper presents a new inductive learning algorithm, HGR (Version 2.0), based on the newly-developed extension matrix theory The basic idea is to partition the positive examples of a specific class in a given example set into consistent groups, and each group corresponds to a consistent rule which covers all the examples in this group and none of the negative examples. Then a performance comparison of the HGR algorithm with other inductive algorithms, such as C4.5,OC1, HCV and SVM, is given in the paper. The authors not only selected 15 databases from the famous UCI machine learning repository, but also considered a real world problem. Experimental results show that their method achieves higher accuracy and fewer rules as compared with other algorithms.     

基于超立方体覆盖的构造性网络学习算法

该文提出了一种基于超立方体覆盖的构造性神经网络学习算法,以解决二值型输入变量的K分类问题。该算法分两步来动态地构造一个三层前馈网络。首先,对于每一类的所有训练样本,用尽可能少的超立方体来覆盖它们,并为每一个超立方体构造一个隐层单元;其次,用"或"操作把这些隐单元连接到相应的输出单元上。文章给出了相应的理论分析和一个具体的实现。实验结果表明,该算法优于常用的一些归纳学习算法。